Graduate Student Profile - Zhongyin John Daye

Written by: Meghan Honerlaw, M.S. candidate in Statistics

Zhongyin John DayeZhongyin John Daye
Zhongyin John Daye is a fifth year statistics Ph.D. student. He was born in Xi’an, China and grew up near Pasadena, California. His father, a professor in Mathematics and Mechanics at Jiaotong University, was invited to U.S. as a visiting scholar in 1981 by Y.C. Fung, founder of Biomechanics and father of Bioengineering. Following in his father’s footsteps, John first studied Bioengineering at University of California, San Diego. Later, he also studied Electrical Engineering and Computer Science at University of California, Berkeley and Financial Engineering at Columbia University.

After studying these various subjects, John became "impressed with the way statistics could be applied to many different areas of the sciences and its emphasis on both fundamental issues and real-world problems." Besides allowing him to ‘play in everyone’s backyard’, John was intrigued that statistics was a relatively young field and "could be more susceptible to paradigm change."

When John realized he was interested in studying statistics, he sought a university that had a "reputable department with diverse faculty interests and a congenial environment for graduate study." He found that Purdue exemplified all of these traits. Since coming to Purdue, John has taken advantage of the excellent curriculum that the Department offers. He says he "especially likes the seminars and topic courses, which offer opportunities to learn about current research from top experts." John has found that the department has a rich and active research environment. Not only does the department provide a strong technical education, but it promotes "creativity, independent thinking, and integrity in research."

John has had the opportunity to develop these skills while working on several projects. In a joint work with Xinge Jessie Jeng, a fellow Ph.D. student in Statistics at Purdue, they proposed the weighted fusion, a new variable selection method for data with correlated variables. John explains, "In high-dimensionality, data often exhibit multicollinearity. Under this difficult setting, many model selection techniques may fail. Weighted fusion is special in performing model averaging via penalized regression and can often outperform current methods for highly correlated data." John and Jessie gave an invited talk on their completed work at the First Midwest Statistics Research Colloquium held in Chicago on March 28-29, 2008.

Currently, John is pursuing additional topics in model selection with his advisor, Professor Michael Zhu. They are working on developing computational techniques and theoretical results for current model selection techniques as well as designing new methodologies for real-world applications. John says, "We are developing a new methodology for model selection under prior knowledge in the linear setting. Both theoretical results and real-data applications will be presented. In addition, as many data sets may not be adequately described by linear models, we are developing new methodologies and computational algorithms for model selection in the non-linear setting." With his advisor and collaborators, John hopes to make fundamental contributions in the area of model selection and to take on leadership roles in some aspects of its development.

During the past four years at Purdue, John has realized, "Whether one hopes to pursue a career in academia or industry, Purdue offers an invaluable opportunity to learn about research and to build lasting relationships with both faculty and students. Many Purdue alumni have gone on to become leaders in both academia and industry."

July 2008